This page is an accompanying page for the thesis A Data-driven Bayesian Approach to Automatic Rhythm Analysis of Indian art Music by Ajay Srinivasamurthy. Please see http://compmusic.upf.edu/phd-thesis-ajay for more details on the thesis.

The page lists some audio examples of automatic meter analysis in both Carnatic and Hindustani music. Click on each example to expand and listen to outputs of different meter analysis algorithms.

The audio examples have audible clicks marking the different events of the tala. Each track has a high frequency (2000 Hz) loud click at the sama, a lower frequency (1500 Hz) click at the anga/vibhaag boundaries, and a low frequency (1000 Hz) click at each beat of the tala.

The abbreviations correspond to the different algorithms presented in the thesis - Ground Truth: Manually annotated beats, Inf-AMPFo: Meter Inference with AMPF0, Track-AMPFo: Meter tracking with AMPF0, Track-AMPFm: Meter tracking with AMPFm, Track-AMPFs: Meter tracking with AMPFs, tInfTrack-AMPFo: Tempo-informed meter tracking with AMPF0, tInfTrack-AMPFs: Tempo-informed meter tracking with AMPFs, tsInfTrack-AMPFo: Tempo-sama-informed meter tracking with AMPF0, tsInfTrack-AMPFs: Tempo-sama-informed meter tracking with AMPFs.

The Carnatic examples are from CMR dataset

The Hindustani examples are from HMRs and HMRl datasets